Research map / Vision-language and multimodal / Robot action and navigation
Speaker-Follower Models for Vision-and-Language Navigation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add speaker-generated route-instruction pairs for training (ablation) | Baseline follower · R2R val-unseen | SR: 19.9 -> 24.6 | Helped |
| add speaker-generated route-instruction pairs for training (ablation) | Baseline follower · R2R val-seen | SR: 40.3 -> 46.8 | Helped |
| use all augmented data rather than 25% (ablation) | Speaker-Follower · R2R val-unseen | SR: 22.8 -> 24.6 | Helped |
| add speaker-based pragmatic rescoring of candidate routes (ablation) | Data-augmented follower · R2R val-unseen | SR: 24.6 -> 39.3 | Helped |
| add speaker-based pragmatic rescoring of candidate routes (ablation) | Data-augmented follower · R2R val-seen | SR: 46.8 -> 57.2 | Helped |
| use speaker scores instead of only follower scores to select candidate routes (ablation) | Speaker-Follower search · R2R val-unseen | SR: 30.5 -> 39.3 | Helped |
| replace low-level visuomotor control with panoramic actions (ablation) | Data-augmented follower with pragmatic inference · R2R val-unseen | SR: 39.3 -> 54.6 | Helped |
| remove speaker-generated training pairs (ablation) | Full model · R2R val-unseen | SR: 54.6 -> 49.5 | Hurt |
| remove speaker-based route rescoring (ablation) | Full model · R2R val-unseen | SR: 54.6 -> 35.5 | Hurt |
| replace panoramic actions with non-panoramic control (ablation) | Full model · R2R val-unseen | SR: 54.6 -> 39.3 | Hurt |
| use the challenge participation sequential inference procedure (ablation) | Speaker-Follower · R2R test (unseen) | OSR: 63.9 -> 96.0 | Helped |
| use state-factored rather than standard beam search (ablation) | Speaker-Follower pragmatic inference · R2R val-unseen | SR: 50.7 -> 54.6 | Helped |
Closest papers by meaning
- Counterfactual Cycle-Consistent Learning for Instruction Following and Generation in Vision-Language Navigation (CVPR 2022 · similarity 0.91)
- Aux-Think: Exploring Reasoning Strategies for Data-Efficient Vision-Language Navigation (NeurIPS 2025 · similarity 0.91)
- Instruction Decomposition and Action Alignment for Vision-Language Navigation (ICML 2026 · similarity 0.90)
- NavGPT: Explicit Reasoning in Vision-and-Language Navigation with Large Language Models (AAAI 2024 · similarity 0.89)
- NavBench: Probing Multimodal Large Language Models for Embodied Navigation (NeurIPS 2025 · similarity 0.88)
- Run, Ruminate, and Regulate: A Dual-process Thinking System for Vision-and-Language Navigation (AAAI 2026 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.